爬取高考网历年分数线并可视化

萧易天

1. import requests
   
2. from lxml import etree
   
3. from pyecharts import options
   
4. from pyecharts.charts import Bar
   
5. from pypinyin import lazy_pinyin
   
6. def get(url):
   
7.     response = requests.get(url).content.decode('gbk')
   
8.     #print(response)
   
9.     html = etree.HTML(response)
   
10.     year=html.xpath('//th[@width="72"]/text()')[0:11]
    
11.     wenke=html.xpath('//tr[@class="c_blue"]/td/text()')[0:12]+html.xpath('//tr[@class="c_white"]/td/text()')[0:12]
    
12.     wen = []
    
13.     for i in wenke:
    
14.         i = i.replace('\r','').replace('\n','').replace('\t','')
    
15.         wen.append(i)
    
16.     like = html.xpath('//div[@class="cjArea tm15"]/table[last()]/tr[@class="c_blue"]/td/text()')[0:12]+html.xpath('//div[@class="cjArea tm15"]/table[last()]/tr[@class="c_white"]/td/text()')[0:12]
    
17.     li = []
    
18.     for i in like:
    
19.         i = i.replace('\r', '').replace('\n', '').replace('\t', '')
    
20.         li.append(i)
    
21.     visulize(pro,li,wen,year)
    
22.     pass
    
23. def visulize(pro,li,wen,year):
    
24.     wen1 = wen[1:12]
    
25.     wen2 = wen[13:24]
    
26.     li1 = li[1:12]
    
27.     li2 = li[13:24]
    
28.     c = Bar()  #生成条形图
    
29.     #添加x，y
    
30.     c.add_xaxis(year)
    
31.     c.add_yaxis('文科一本',wen1)
    
32.     c.add_yaxis('文科二本', wen2)
    
33.     c.add_yaxis('理科一本', li1)
    
34.     c.add_yaxis('理科二本', li2)
    
35.     c.set_global_opts(title_opts=options.TitleOpts(title=pro+"历年高考分数线",subtitle="2009-2019年"))
    
36.     c.render(pro+"历年高考分数线.html")
    
37.     pass
    
38. if __name__ == '__main__':
    
39.     pro = input('想要获得的省份：')
    
40.     prov = lazy_pinyin(pro)
    
41.     prov = "".join(prov)
    
42.     url = 'http://www.gaokao.com/'+prov+'/fsx/'
    
43.     get(url)

复制代码

海月清辉

我估计这是离我最近的可视化数据了

海月清辉

1. 想要获得的省份：湖南
   
2. C:\Users\Administrator\AppData\Local\Programs\Python\Python38\lib\site-packages\pyecharts\charts\chart.py:14: PendingDeprecationWarning: pyecharts 所有图表类型将在 v1.9.0 版本开始强制使用 ChartItem 进行数据项配置 :)
   
3.   super().__init__(init_opts=init_opts)
   
4. 
   
5. Process finished with exit code 0
   

复制代码

leizi666

抱错呀，跑不起来